[time-nuts] Z3801A Frequency Jump
Magnus Danielson
magnus at rubidium.dyndns.org
Wed Jan 27 08:11:20 UTC 2010
SAIDJACK at aol.com wrote:
> Hi Brian,
>
> This is a classical crystal jump.
>
> Nothing one can do about it.
>
> Could have been internal to the crystal (stress-relief) or external (gamma
> particle hitting the crystal lattice etc).
>
> My 58503A does it every couple of days or so. Really messes with your ADEV
> performance...
Actually, it messes up the calculated ADEV if the frequency jump is not
separated from the noise. You could argue for keeping the jump in or out
of the ADEV estimation, One should notice that the effect on the ADEV
estimate of the jump varies with the length of the sample-series and the
estimator being used, so the argument that one want it "in" to see the
effect on ADEV is in my opinion a vauge one. A frequency jump should be
separated out and show separated as just that, it's a problem in its own
right. If you want it "in" the ADEV you need to have a sample series
hold about 10 of them or so to make meaningfull statistics...
Cheers,
Magnus
More information about the time-nuts
mailing list